The proposed research combines theoretical and experimental investigations of the neural basis of olfactory recognition in the rat. A proposed computational model uses spike synchrony to implement an algorithm capable of concentration- and background-invariant odor recognition. The model is consistent with the known anatomy and physiology of the olfactory bulb and provides a formal substrate that links predictions and mechanisms from the neurophysiological to the behavioral levels. From this model arise three specific aims for the current funding period: Specific Aim 1 is to test the hypothesis that odors are encoded by the relative degree of activation of different glomeruli, not merely by the binary spatial pattern of which glorneruli are activated above a fixed threshold. In the model, logarithmic glomerular encoding using analog values facilitates concentration- invariant recognition. [unreadable] In the proposed model, synchronization can be used to integrate odor representations distributed across the olfactory bulb; the identity of the set of mitral cells that synchronize specifies the odor. Specific Aim 2 is to test the hypothesis that synchrony of mitral cell spikes in the mammalian olfactory bulb is odor-specific and thus a candidate odor representation. Each glomerulus contains the primary dendrites of many mitral cells, yet it encodes the activation of a single type of olfactory receptor. Specific Aim 3 is to test the hypothesis that different mitral cells receiving the same glomerular input respond differentially to this input. In the proposed model, this provides a substrate for odor recognition templates. The experimental techniques used to test these hypotheses range from intrinsic optical imaging and whole-cell recording in anesthetized animals to microstimulation and electrophysiological recording during psychometric odor discrimination tasks performed by behaving animals. Throughout this investigation, there will be reciprocal interactions between these experimental approaches and computational modeling. If successful, this work will provide a detailed computational account of the neural implementation of key computational algorithms for the process of odor recognition. [unreadable] [unreadable]